<![CDATA[Ensuring the safety and health of fish products is crucial for public health, with tuna being Indonesia's second most popular fishery product. Tuna freshness is a key indicator of seafood safety, directly impacting both nutritional quality and contamination risk. This study compares the K-Nearest Neighbors (KNN), Naive Bayes, and Support Vector Machine (SVM) algorithms to assess and classify tuna freshness, offering an accurate and efficient approach. A machine learning model categorized Tuna freshness based on the gases emitted, utilizing a dataset of 58,389 records. Gas changes were detected using the MQ-135, MQ-9, and MQ-2 sensors, which are highly sensitive to gases like ammonia, methane, and alcohol, commonly associated with spoilage. The KNN, Naive Bayes, and SVM algorithms were then applied to classify the sensor data. KNN and SVM achieved an accuracy of 99%, while Naive Bayes reached 90%. The high accuracy of these methods highlights their potential as practical tools for the fishing industry, enabling suppliers and retailers to assess tuna freshness more effectively. This method could significantly improve consumer safety by ensuring only high-quality, fresh products reach the market. Additionally, automation offers substantial time savings, facilitates faster decision-making, and reduces reliance on manual inspections prone to human error. Keywords—tuna, classification, gas sensor, machine learning]]>
Copyrights © 2025
